> > any significant performance increase in any case.
>
> I disagree.
>
> My tests:
>
> the function queryRandomize() from DRK 4 (or was it 5?), I
> rewrote it, line-by-line (including comments!) and called it
> queryRandomizeCFML().
>
> 100,000 iterations of each try, randomizing a 1 column query
> with 20 rows, time in ms, server is CFMX 6.1 standalone
> developer, 3 tries:
>
> script tags
> 14640 26342
> 14718 25983
> 14609 26108
I submit that your test provides insufficient data to justify that
conclusion.
First, running a single algorithm serially within a loop doesn't provide an
especially useful measurement of how that algorithm will work when executed
in parallel. I've seen all sorts of things that seem faster when measured
within a loop, but don't seem faster when measured by load testing - I've
even seen things that are slower when measured by load testing, even though
they were faster within a loop!
Second, you'd need to test more than one algorithm to contradict the general
statement about CFSCRIPT not providing any significant performance increase.
It may run faster when manipulating a query object, for example, but may be
slower for everything else.
Third, you're pointing out an approximate difference of three milliseconds
per request, which is practically immeasurable by CF (if I recall correctly,
CF can't accurately handle time increments smaller than 10 milliseconds). I
would question whether that's a significant performance increase, especially
given the second point above.
Finally, as with all these "which is faster" questions that come up, even if
you are correct in your supposition, the time savings you may get by doing
this is typically dwarfed by that gained by additional caching. I don't
think I've ever seen an application which has taken full advantage of every
caching opportunity. So, the question becomes whether you want to spend your
time optimizing your code based on your conception of what is faster within
a given version of CF, or whether you want to spend your time making your
application faster no matter what version of CF you use.
Dave Watts, CTO, Fig Leaf Software
http://www.figleaf.com/
phone: 202-797-5496
fax: 202-797-5444
[Todays Threads] [This Message] [Subscription] [Fast Unsubscribe] [User Settings]
